J.E. Opfer / Cognition 86 (2002) 97–122
97
Cognition 86 (2002) 97–122
www.elsevier.com/locate/cognit
Identifying living and sentient kinds from dynamic
information: the case of goal-directed versus aimless
autonomous movement in conceptual change
John E. Opfer
Department of Psychology, Carnegie Mellon University, Baker Hall 331B, Pittsburgh, PA 15213, USA
Received 5 September 2001; received in revised form 4 July 2002; accepted 9 August 2002
Abstract
To reason competently about novel entities, people must discover whether the entity is alive and/or
sentient. Exactly how people make this discovery is unknown, although past researchers have
proposed that young children – unlike adults – rely chiefly on whether the object can move itself.
This study examined the effect of goal-directed versus aimless autonomous movement on children’s
and adults’ attributions of biological and psychological capacities in an effort to test whether goaldirectedness affects inferences across documented periods of change in biological reasoning. Half of
the participants (adults, and 4-, 5-, 7-, and 10-year-olds; Ns ¼ 32) were shown videos of unfamiliar
blobs moving independently and aimlessly, and the other half were shown videos of identical blobs
moving identically but toward a goal. No age group was likely to attribute biological or psychological
capacities to the aimless self-moving blobs. However, for 5-year-olds through adults, goal-directed
movement reliably elicited life judgments, and it elicited more biological and psychological attributions overall. Adults differed from children in that goal-directed movement affected their attributions
of biological properties more than their attributions of psychological properties. The results suggest
that both young children and adults consider the capacity for goal-directed movement to be a decisive
factor in determining whether something unfamiliar is alive, though other factors may be important in
deciding whether the thing is sentient. q 2002 Elsevier Science B.V. All rights reserved.
Keywords: Categorization; Conceptual change; Goal-directed movement; Animacy; Naive biology; Intentionality; Teleology; Cognitive development
1. Introduction
Since Piaget (1929), many psychologists have argued that people of various ages
identify things as alive or sentient by seeing these things move on their own (Bassili,
E-mail address: [email protected] (J.E. Opfer).
0010-0277/02/$ - see front matter q 2002 Elsevier Science B.V. All rights reserved.
PII: S 0010-027 7(02)00171-3
98
J.E. Opfer / Cognition 86 (2002) 97–122
1976; Laurendeau & Pinard, 1962; Mandler, 1992; Poulin-Dubois, Lepage, & Ferland,
1996; Premack, 1990; Reed, 1997; Stewart, 1982). This insight no doubt captures an
important phenomenon: a child who sees her once active hamster lying still and unresponsive will be horrified, not because she thinks the inaction to be a temporary state, but
because for her it signals that the hamster is now and forever dead and unaware.
Whether something moves by itself, however, may be anything but clear-cut strictly at
the perceptual level. Leaves, paper cups, cars, and airplanes often move without any
apparent external force, yet adults are not tempted to identify them as living things. For
the leaves and paper cups, adults might infer that the wind causes them to move, and for
the cars and airplanes, that some inanimate motor does the work. In both cases, adults
consider these moving things as inanimate and insentient as a dead hamster.
Adults’ interpretations of movement, of course, may be influenced by their prior knowledge: they know that leaves and paper cups have no internal mechanism for locomotion,
and they know that cars and airplanes move only because a person starts their motor. If
children say otherwise, is it not fair to assume that they merely lack the knowledge that
adults possess? That is, if confronted with an unfamiliar object moving autonomously,
might not children and adults both infer that the self-moving object is a living thing?
Generally, they do not: even when entities are unfamiliar, children and adults do not
consider autonomous movement to be a decisive factor in judging something alive or
sentient (Gelman, Durgin, & Kaufman, 1995; Poulin-Dubois & Héroux, 1994; Richards &
Siegler, 1986). For example, when asked to identify an object that moved with multiple
accelerations without any apparent outside force, one adult responded “(long pause) Gosh,
I don’t know … let’s try just a windblown object. You’re in an area where it is swirling and
stopping and changing direction…” (Gelman et al., 1995, p. 168). Based on such
responses, Gelman et al. (1995) concluded, “Animates cannot be distinguished from
inanimates simply on the basis of motion because the cues of motion, like the static
cues of color and shape, are ambiguous” (p. 156).
This conclusion raises several important questions, questions that concern not only
adults’ conceptions of animacy, but also questions about the role of movement in the
origin, development, and character of that understanding. First, can animates be distinguished from inanimates simply based on motion? Might not some sort of movement (such
as goal-directed movement) independently and unambiguously signal whether a novel
object is an animate? Second, does the meaning of different types of movement change
over the lifespan? For example, do young preschoolers through adults decide whether
something is alive by reference to the same standard (such as whether it can move itself
toward goals), or do they use incommensurable standards for deciding whether something
is alive? Lastly, what exactly is the meaning of different categories of movements? That is,
if there are unambiguous motion categories that imply animacy, do they directly signal life
or something else that implies life (such as sentience or intentionality, i.e. the capacity to
sense, represent, or desire goals)?
The present experiment seeks to address these questions by asking preschoolers through
adults about novel entities that move in two distinct ways – ones that move autonomously
but aimlessly and ones that move autonomously but toward a goal. Goal-directed movement, I propose, unambiguously signals animacy where autonomous movement fails to do
so. I argue that this goal-directed movement is a commensurate standard by which
J.E. Opfer / Cognition 86 (2002) 97–122
99
preschoolers through adults identify novel entities as living things, that its value for
identifying living things operates even during periods of profound conceptual change in
the biological domain, and that this use of goal-directed movement may make many of
these conceptual changes possible.
1.1. The ambiguity of autonomous movement
Autonomous movement is a type of movement in which an entity moves itself, with
independent changes in direction and acceleration. Because this movement is so characteristic of living things, many researchers have thought it powerful enough to motivate
young children’s animacy–inanimacy distinction (see Gelman & Opfer, in press, for
review). However, many empirical attempts have failed to confirm this intuition. For
familiar entities, autonomous movement has proven incidental to the other sorts of information that children use to identify animates (e.g. natural kind status; Bullock, 1985;
Dolgin & Behrend, 1984; Richards & Siegler, 1984). For unfamiliar entities, the movement provides only ambiguous biological information (Gelman et al., 1995; PoulinDubois & Héroux, 1994; Richards & Siegler, 1986).
Some of the earliest experiments on the biological interpretation of unfamiliar entities
were performed by Stewart (1982, 1984) in a series of unpublished studies with adults.
Stewart proposed that any non-Newtonian motion, including autonomous motion, would
yield an illusion of animacy. To test her proposal, she generated computer representations
of a ball that systematically differed in its movement: (1) the ball moved either as a result
of collision with another ball or with no visible source of initiation; (2) either the ball
collided with another surface and changed directions or it changed directions before
colliding with another surface; and (3) either the ball moved at a constant speed or the
ball doubled its speed midpath without any apparent outside force. Gelman et al. (1995)
recently repeated Stewart’s methodology. For the first pair of movements, participants
were more likely to judge the self-propelled ball as alive (25%) than the externally
propelled one (0%). For the second pair, the ball that ‘avoided’ the barrier was also
more likely to be judged alive (12.5%) than the ball that ‘bounced off’ the barrier (0%).
For the third pair, both balls were judged not alive by roughly 60% of the participants. In
conjunction, these studies show that autonomous motion provides adults with weak but
uncompelling animacy cues.
Children’s judgments of unfamiliar objects have been similar. For example, PoulinDubois and Héroux (1994) presented 5-, 7-, and 9-year-olds with computer displays of
unfamiliar, irregularly-shaped stimuli that moved either autonomously or because of an
external agent. Children were then asked whether the entity was “alive” and whether it
possessed other biological and psychological properties. They found that autonomous
motion was more likely than non-autonomous motion to elicit such attributions, but
again neither movement elicited more than 4.5 of the 12 possible attributions of animate
objects, a level indistinguishable from random guessing. These results led the authors to
allow for the same conclusion reached by Gelman et al. (1995): “[E]ven for 5-year-old
children, other features must also be involved [in attributing life] as their low scores for the
artificial stimuli indicate” (p. 24). Given that autonomous motion seemed an ambiguous
animacy cue, it is worth examining what “other features” might provide a stronger cue.
100
J.E. Opfer / Cognition 86 (2002) 97–122
1.2. Goal-directed versus aimless autonomous movement
Another feature that might indicate life (Opfer & Gelman, 2001; Opfer & Siegler, 2001)
or sentience (Csibra & Gergely, 1998) is goal-directedness. Goal-directed movement is a
type of autonomous movement in which the agent contingently directs its movement
toward (or away from) another object, state, or location. ‘Movement’ is used here in the
full sense used by Michotte (1963), which includes “changes in shape”, which he classified
as the special movement of the “kinaesthetic amoeba” (pp. 204–206). Thus, a plant
growing toward the sun or an amoeba engulfing a paramecium would equally count as
movements. ‘Goal’ refers here to any object, state, or location toward which an entity
contingently directs its self-movement. Thus, a ceiling fan (say) would not be the goal of a
plant’s growth (even if a plant grew toward it) because the plant would not grow toward
the ceiling fan were it moved elsewhere; in this case, the growth is only accidentally
directed toward the ceiling fan. In contrast, the goal of a plant’s growth could be a stream
of sunlight. For example, imagine that sunlight is streaming from the left and right sides of
a shaded window. Place a plant on the right side of the window and it will consistently
grow leftward (i.e. toward the sunlight); move the plant to the left side of the window, and
it does not continue to grow leftward (i.e. away from the sunlight) but instead changes the
direction of its growth to the right (i.e. toward the sunlight). The difference between
accidental self-movement toward objects (such as ceiling fans) and contingent self-movement toward objects (such as sunlight) is the difference that ‘goal-directed movement’ is
meant to convey.
Goal-directed movement is superficially similar to two types of contingent, objectdirected movement that are not autonomous – being pushed toward a goal and getting
pulled by another object. An object (O) is readily perceived as pushed by another object
(S) when there is contact between S and O, when the speed of S is at least half as great as
the post-contact speed of O, when the movement of O immediately occurs upon contact,
and when the direction of the post-contact movement of O is similar to the direction of the
pre-contact movement of S (Michotte, 1963). Thus, if O moves to a goal (G) following
Michotte-like pushing, the movement of O cannot be classified as goal-directed (though it
might result from the goal-directed movement of S). For example, if Michael Jordan
consistently hits his baskets, we would say that the movements of Michael Jordan were
goal-directed, but we would not say that the movement of the basketball itself was goaldirected. Indeed, 3- and 5-year-olds reliably see the difference between being pushed
toward a goal and moving there autonomously, and 5-year-olds do not claim that a person
pushed toward a goal actually desires it (Montgomery, 1996).
A similar type of object-directed (though not goal-directed) movement is pulling (White
& Milne, 1997). Adults’ impression of pulling arises from seeing one object following
another, even when there is no visible connection between the pulling and pulled objects
and the two objects move in different planes. Their impression of pulling, however, is
decreased (and their impression of autonomy increased) by two factors: (1) delay before
movement, wherein a delay precedes one object following a second object; and (2) change
in direction, wherein the first object moves in the opposite direction as the second object
moves, and only then follows the second object (White & Milne, 1997).
Although no studies have systematically varied the factors that give rise to impressions
J.E. Opfer / Cognition 86 (2002) 97–122
101
of goal-directedness, it seems plausible that the features that attenuate impressions of
pushing and pulling amplify impressions of goal-directedness if for no other reason
than that they also signal autonomy. The delay before movement may signal that a
spontaneous change within the agent is necessary for it to move toward its goal. For
example, an amoeba needs time for the chemicals emitted by a paramecium to reach it
before the amoeba moves toward it, and a person needs a moment at least to identify a cup
of coffee before reaching for it. The change in direction highlights that the agent is not
already moving in a particular direction, but perhaps that the goal (or the idea of the goal)
itself affected some change in the agent. For example, if a person steps on her cat’s tail on
the way to the refrigerator, we might say that stepping on the tail was accidental rather than
goal-directed. If, however, the cat then ran away from the person’s path to the refrigerator
and the person immediately changed her direction and again stepped on her cat’s tail, we
would then doubt the tail-stepping to be accidental.
Whether delay before movement and change in direction, either independently or in
combination, or other factors actually determine impressions of goal-directedness is
unclear. Some authors, for example, emphasize the importance of change in direction
(Poulin-Dubois & Héroux, 1994; Premack & Premack, 1997) and delay before movement
(Poulin-Dubois & Héroux, 1994). Others claim that an agent must additionally terminate
its action upon reaching its goal (satisfaction) (Premack & Premack, 1997) and that it must
move directly toward the goal (direct trajectory) (Gergely, Nadasdy, Csibra, & Biro, 1995;
Premack & Premack, 1997).
In light of these features, it is worth reconsidering one study suggesting that goaldirected movement does not elicit any more life and sentience judgments from children
and adults than aimless autonomous movement. Richards and Siegler (1986) presented a
computer image of a rectangle to 5- to 6-year-olds. The children were told that they were
to imagine that they had gone to another planet in search of living things and “sometimes
you’ll put [a toy truck] in front of the object to see if the object goes to it”. In one
condition, the rectangle moved without any apparent external impetus (autonomous
motion). In another condition, the rectangle moved to the depicted toy truck (goal-directed
motion). No age group, either alone or in combination with other factors (such as whether
the rectangle had legs or wheels, and whether the terrain was flat, uphill, or downhill),
associated this goal-directed activity with life. The authors concluded that either the toy
truck did not represent a goal to the children or that goal-directed movement was not
important for inferring that the agent was alive.
Referring back to the proposed signals of goal-directedness, there are at least three
reasons to agree with Richards and Siegler’s suggestion that the children had no cause
to interpret the toy truck as a goal for the alien life forms. First, there was no delay before
the rectangle moved to the toy truck. A plausible interpretation, then, might be that the
rectangle was pulled by a magnet. Second, the rectangles were not shown changing
direction toward the toy truck; therefore, it was impossible to determine whether the
rectangle’s movement to the toy truck was accidental or goal-directed. Third, the scenario
did not specify what would satisfy the rectangle, that is, indicate the conditions under
which the rectangles do or do not pursue goals. Therefore, the toy truck may have seemed
an unlikely goal for the rectangles.
In sum, neither attributions of sentience nor life have been shown to rest on seeing an
102
J.E. Opfer / Cognition 86 (2002) 97–122
object move autonomously or toward goals, but these failures seem to stem from different
causes. Displays of autonomous motion have played little role in biological and psychological judgments seemingly because autonomous motion itself is ambiguous (Gelman et
al., 1995; Poulin-Dubois & Héroux, 1994). The display of goal-directed motion, however,
seems to have had little effect because the particular presentation of goal-directed movement did not vary important features of goal-directedness (e.g. delay before movement,
change in direction, and indication of satisfaction).
1.3. Understanding goal-directedness as a biological versus psychological capacity
Preschoolers’ previous failures to base their judgments on movement type do not appear
to have resulted from their age: even adults failed to attribute biological and psychological
properties to the stimuli heretofore described, possibly because the displays themselves
were ambiguous rather than because adults cannot identify animates on the basis of movement. Thus, if we eliminate the ambiguity of these stimuli and we find that adults do
differentiate autonomous and goal-directed motion, then we have reason to believe that
preschoolers may do so as well. Such findings would then raise the question: when goaldirectedness signals animacy, does it primarily elicit biological or psychological judgments? That is, does a thing’s goal-directed movement imply that it has psychological
capacities (such as the ability to sense or desire the goal), just biological capacities (such as
the ability to eat the goal), both, or neither? Which meaning people take from goaldirected movement at various ages is important because it helps to resolve issues about
the nature of conceptual change in the biological domain.
First, goal-directed movement may imply neither biological nor psychological capacities. A non-mentalistic, non-biological interpretation of goal-directed movement has been
attributed to infants participating in several studies (Csibra & Gergely, 1998; Csibra,
Gergely, Biro, Koos, & Brockbank, 1999; Gergely & Csibra, 1997; Gergely et al.,
1995). Young children, too, may attribute goals to agents without spontaneously identifying the mental or physiological causes of such actions or the psychological or biological
significance of such goals. Under this proposal, a young child faced with goal-directed
movement is concerned chiefly with the determination of goal-directedness per se and is
otherwise agnostic about the agents’ biological and psychological properties, including
even its ability to move autonomously.
Second, young children may initially predicate “life” and other biological capacities on
the capacity to act intentionally (which is psychologically-caused goal-directed action),
thereby leading them to wrongly exclude plants and other insentient life forms (such as
germs) from their living things concept and biological reasoning (Carey, 1985; Solomon &
Cassimatis, 1999). Thus, until children change their minds about the life-intention connection, they cannot understand why organisms they know to be insentient (e.g. plants) are
“alive” (Carey, 2000). In this view, knowing that something acts toward goals elicits
preschoolers’ judgment of life only if they also judge the action intentional. If this characterization of preschoolers’ beliefs is correct, then showing them a novel entity selfmoving toward a goal should affect their biological and psychological judgments equally
and preschoolers’ judgment that something is alive should be no more likely than their
judgment that it is psychological.
J.E. Opfer / Cognition 86 (2002) 97–122
103
Lastly, children may predicate “life”, not on the capacity for intentional action, but on
the capacity for goal-directed action, whether presumed intentional or not. If children
equate life with the capacity to act toward goals, then no change in the concept “life” is
necessary to understand why plants and other insentient organisms are alive: they need
only learn that plants and other insentient organisms act toward goals, just as animals must
do to be alive. According to this characterization, knowing that a thing acts toward goals is
sufficient to know that it is alive, whether or not the action is believed to be intentional. A
stronger (though logically unnecessary) version of this view is that preschoolers understand that goal-directed actions may be living actions yet non-intentional (Keil, 1992). If
this strong view is correct, then evidence of goal-directedness should increase the likelihood of biological judgments more than psychological ones, as opposed to increasing
them equally (according to the weaker version).
Interestingly, biological versus psychological interpretations of goal-directed action
seem to change between the ages of 5 and 10 years. Opfer and Gelman (2001), for
example, investigated the sorts of information 5-year-olds through adults use to predict
and explain teleological action (i.e. self-beneficial, goal-directed movement). Participants
were asked whether animals, plants, machines, and simple artifacts would direct their
movement toward a needed goal (e.g. an animal toward a mouse, a plant toward sunlight,
or a machine toward electricity) versus toward an object the entity did not need (e.g. an
animal toward a box or a plant and machine toward a picture on the wall). Five-year-olds
seemed to believe that only animals are teleological agents. They were likely to predict
that animals would move toward needed goals, were more likely to predict that animals
would move toward these goals than toward objects they did not need, and were more
likely to predict that animals would act toward goals than that plants and artifacts would.
When explaining their predictions, preschoolers referred to psychological states more
often than any other factor, including biological needs or mechanical forces, and they
attributed psychological capacities only to animals at above-chance levels. From this
evidence, Opfer and Gelman (2001) concluded that preschoolers take goal-directed action
to be a distinctive capacity of psychological living things.
Adults and fifth graders, like preschoolers, restricted their predictions of teleological
action to living things, but they did not seem to think that psychological capacities were
also necessary. For example, both adults and fifth graders predicted that insentient plants
would act in a number of goal-directed ways, such as growing toward sunlight rather than
pictures, growing toward water rather than oil, and enclosing a fly rather than a pebble.
When asked to explain why plants and animals would move toward goals, fifth graders and
adults (unlike preschoolers) were likely to refer to how these goals benefited the actors’
lives, and they were unlikely to attribute psychological capacities to plants (Opfer &
Gelman, 2001). Adults and older children seemed to understand that goals can be pursued
but not desired and that the things that pursue goals are living things (whether mindless or
not).
1.4. The present experiment
Across several studies, children’s and adults’ judgments about the life and sentience of
familiar entities have been only weakly linked to autonomous movement (Bullock, 1985;
104
J.E. Opfer / Cognition 86 (2002) 97–122
Dolgin & Behrend, 1984; Richards & Siegler, 1984). Their expectations of goal-directed
movement, however, have closely followed their judgments of what can be alive (Opfer &
Gelman, 2001) – animals according to 5-year-olds, plants and animals according to 10year-olds and adults (Carey, 1985; Hatano et al., 1993; Richards & Siegler, 1984). These
findings suggest that children and adults, though their judgments of what is alive differ
dramatically, may both base their life judgments on the same dynamic feature – goaldirected movement. Judgments of sentience, in contrast, do not appear to be predicated on
goal-directedness among older children and adults.
The purpose of the present experiment is to test these hypotheses and address the
following questions. (1) Can animates be distinguished from inanimates simply based
on goal-directedness? (2) When in development do children first use goal-directedness
to identify living and sentient kinds? (3) Does the meaning of goal-directedness change
with the development of biological and psychological knowledge?
To answer these questions, this experiment sought to identify the effect of a goal on
children’s and adults’ interpretation of autonomously moving agents. Two sorts of interpretations were of interest here: whether the agent was thought to possess various biological and psychological capacities, and what the agent was identified to be. To rule out
prior biases about the agents’ identity, the agents (blob-like microorganisms) were chosen
based on a pilot study with adults, which established that the blobs were in fact unfamiliar
and not recognizable as microorganisms.
To explore the role that developing biological and psychological knowledge might have
on the interpretation of these goal-directed agents, four age groups of children (4-, 5-, 7-,
and 10-year-olds) were chosen that typically differ greatly in their biological and psychological knowledge. Four-year-olds were selected because they often fail to attribute life
consistently to novel and familiar entities (Carey, 1985; Laurendeau & Pinard, 1962; Looft
& Bartz, 1969; Piaget, 1929; but see Richards & Siegler, 1984), because they have been
said to rely heavily on whether novel agents move independently (Laurendeau & Pinard,
1962; Piaget, 1929), and because they reportedly fail to differentiate the biological and
psychological domains (Carey, 1985). If 4-year-olds – like adults – attribute biological but
not psychological capacities to goal-directed but not autonomous agents, then we would
have evidence consistent with the notion that goal-directed action is a perceptual primitive
in the biological domain.
Five-year-olds were chosen because they attribute life consistently (if incorrectly) to
animals but not plants and artifacts (Dolgin & Behrend, 1984; Richards & Siegler, 1984),
expect that animals will act teleologically (Opfer & Gelman, 2001), and reportedly use
neither autonomous nor goal-directed movement to attribute biological capacities to nonanimals (Dolgin & Behrend, 1984; Poulin-Dubois & Héroux, 1994; Richards & Siegler,
1984). These studies suggest that 5-year-olds reason about animals in a uniquely biological manner and that they are not willing to reason about other living things (such as
plants and germs) similarly (Carey, 1985).
The 7- and 10-year-olds were selected because their knowledge of the biological and
psychological domains has been characterized as adult-like in most of the respects relevant
here (Carey, 1995). Neither of these age groups has been shown to reliably attribute
biological properties to autonomously moving things (Poulin-Dubois & Héroux, 1994),
J.E. Opfer / Cognition 86 (2002) 97–122
105
and at least 10-year-olds expect that both insentient plants and sentient animals will act
teleologically (Opfer & Gelman, 2001).
Adults were included because they presumably know of many more insentient living
things than do children. Accordingly, even if goal-directedness is an equally strong cue
across all ages that something is a living thing, the effect of the cue on adults’ psychological attributions should still be less than that on children’s attributions. In this way,
developing knowledge about the scope of living things could yield different sorts of
interpretations about goal-directed agents: for children, goal-directedness would serve
as an equally strong biological and psychological cue, whereas for adults, goal-directedness would serve as a stronger biological than psychological cue. Similarly, as children
learn about what is necessary to be a living thing (e.g. dying, reproducing, needing water
and nutrients), goal-directedness should have a successively greater impact on biological
attributions generally.
2. Method
2.1. Participants
Participants included 4-year-olds (Goal condition: N ¼ 16, ages 4.01–4.87, mean
age ¼ 4:51; No Goal condition: N ¼ 16, ages 4.01–4.99, mean age ¼ 4:48), 5-year-olds
(Goal: N ¼ 16, ages 5.09–5.93, mean age ¼ 5:41; No Goal: N ¼ 16, ages 5.03–5.98,
mean age ¼ 5:46), 7-year-olds (Goal: N ¼ 16, ages 7.33–8.44, mean age ¼ 7:84; No
Goal: N ¼ 16, ages 7.51–8.37, mean age ¼ 7:91), 10-year-olds (Goal: N ¼ 16, ages
10.66–11.22, mean age ¼ 10:92; No Goal: N ¼ 16, ages 10.41–11.21, mean
age ¼ 10:71), and adults (Goal: N ¼ 16, ages 18.69–21.44, mean age ¼ 19:62; No
Goal: N ¼ 16, ages 18.18–21.45, mean age ¼ 19:56). Four- and 5-year-olds were
recruited from a university-affiliated preschool, and 7- and 10-year-olds were recruited
from a suburban public elementary school. Adults volunteered to fulfill their introductory
psychology class research participation requirement.
2.2. Materials
The materials comprised eight QuickTimee videos of moving blobs; half of these
videos depicted a blob moving itself toward a goal (an irregularly-shaped dark dot), and
the other half were identical except that the goal was removed. All the agents themselves
were irregularly-shaped, such that the agent could be viewed as having an orientation. The
paths of the agents’ and goals’ movement are depicted in Fig. 1.
Each of the videos depicted at least one of the properties thought to indicate goaldirected movement: goal-directed change in the agent’s orientation (e.g. pointed left
toward the goal and then right toward the goal), goal-directed change in the direction of
the agent’s movement (e.g. moving toward the goal to the left and then toward the goal to
the right), delay before the agent’s movement (i.e. moving only when the goal is present),
and change in movement upon reaching the goal (e.g. ceasing movement upon reaching
the goal or changing direction toward a new goal). One change in orientation was depicted
in Videos A, B, and D, and three in Video C. One change in direction was depicted in
106
J.E. Opfer / Cognition 86 (2002) 97–122
Fig. 1. Video presentations of agent and goal paths: Goal condition.
Videos A and B, and three in Video C. A delay before movement was depicted in Videos C
and D. Lastly, changes in movement upon reaching the goal were depicted in Videos A, B,
C, and D.
These videos were selected from a larger group of videos shown to adults in a pilot
study, which sought to ensure that the agents were not identifiable as microorganisms. In
this pilot study, 24 adults were each shown ten stills from a pool of ten videos of microorganisms moving toward no goal. For each still, adults were asked to identify the micro-
J.E. Opfer / Cognition 86 (2002) 97–122
107
organism (1 ¼ living kind, 0 ¼ non-living kind), how confident they were with their
identification (9 ¼ most confident, 1 ¼ least confident), how familiar they were with
those entities (9 ¼ most familiar, 1 ¼ least familiar), whether the entity could move by
itself or whether it required an outside force (0 ¼ externally-caused movement, 1 ¼
autonomous movement), and whether it could move toward something it needs consistently or only accidentally (0 ¼ accidental movement, 1 ¼ goal-directed movement).
Videos A, B, C, and D were the four videos that were most unlikely to be identified as
living kinds (M ¼ 0:16, range: 0.04–0.26) and were the only videos unlikely to be judged
self-moving (M ¼ 0:24, range: 0.17–0.38) and goal-directed (M ¼ 0:19, range: 0.08–
0.25) (Ps , 0:10). Although adults incorrectly identified the microorganisms as nonliving things, they reported moderate familiarity with the items (M ¼ 4:50) though very
low confidence in their judgments (M ¼ 2:80). In short, adults were not sure what the
blobs were, but were more likely to guess that they were clouds, islands, and comets than
to guess that they were organisms that could move autonomously or toward goals.
2.3. Procedure
Participants were randomly assigned to one of two conditions. In the Goal condition,
participants were shown four videos of a blob moving toward a goal (Videos A, B, C, and
D); in the No Goal condition, they were shown the identical videos, except the goal was
removed. Participants were shown only one type of movement (goal-directed or not goaldirected) in order to control for the possibility of contrast effects between the two conditions; however, this design could work against the hypotheses if participants avoid repetitive answers. The order of videos in each condition was counterbalanced across subjects
using a Latin square. After the presentation of each video, participants completed an
attribution task followed by an identification task.
The attribution task presented participants with a battery of questions about the biological and psychological capacities of the agents in the video. The order of biological and
psychological batteries was counterbalanced across subjects, and the order of questions
within each battery was random. The biological battery comprised the following questions: “Can it die?”, “Can it grow?”, “Can it make more little ones just like it?”, “Does it
need something?”, “Does it need food?”, “Does it need water?”, and “Is it alive?”. The
psychological battery comprised these questions: “Can it want something?”, “Can it feel
pain?”, “Can it think?”, “Can it see?”, “Can it be happy?”, “Can it make choices?”, and
“Can it make plans?”.
After the attribution task, participants completed an identification task, in which the
agent in the video was indicated, and participants were asked, “What is it?” In the Goal
condition, participants were also asked “Did this thing [the agent] move so that it could get
closer to this thing [the goal]?” in order to check whether all age groups were equally
likely to judge the entity to be goal-directed. Finally, in both conditions, participants were
asked “Did this thing [the agent] move by itself or did something else move it?” to test
whether judgments of self-movement interacted with the actual type of movements
displayed in the videos.
108
J.E. Opfer / Cognition 86 (2002) 97–122
3. Results
3.1. Attributions of goal-directed and autonomous movement
Each participant in the Goal condition received one score (0–1) indicating the proportion of attributions of goal-directed movement to the four goal-directed agents. A singlefactor (age: 4-year-olds, 5-year-olds, 7-year-olds, 10-year-olds, and adults) ANOVA was
performed on these scores, and post-hoc comparisons used a Newman–Keuls test (unless
otherwise indicated, all post-hoc comparisons used a Newman–Keuls test). There was no
main effect of age, and no age group differed from any other (4s, M ¼ 0:89; 5s, M ¼ 0:94;
7s, M ¼ 0:94; 10s, M ¼ 0:80; adults, M ¼ 0:91). Two-tailed, one-group t-tests indicated
that every age group attributed goal-directed action at above-chance levels (Ps , 0:0001).
In both conditions, participants received one score (0–4) indicating the number of
agents they judged to move autonomously. A 2 (condition: Goal, No Goal) £ 5 (age: 4year-olds, 5-year-olds, 7-year-olds, 10-year-olds, and adults) ANOVA was performed on
these scores. There was a main effect of age (Fð4; 150Þ ¼ 5:42, P , 0:001), indicating that
4-year-olds were less likely to attribute self-movement than 10-year-olds and adults
(Ps , 0:05) and that 5-year-olds were less likely to attribute movement than adults
(P , 0:05). There was also a main effect of condition (Fð1; 150Þ ¼ 13:16, P , 0:001),
indicating that agents in the Goal condition (M ¼ 2:60) were more likely to be judged selfmoving than agents in the No Goal condition (M ¼ 1:89). Age and condition did not
interact.
3.2. Attributions of biological and psychological capacities
In both conditions, each participant received two scores (each ranging from 0 to 7), one
indicating the mean number of biological capacities attributed to the novel agents and
another the mean number of psychological capacities. Thus, if participants guessed
randomly, we would expect 3.5 biological and 3.5 psychological attributions.
A 2 (condition: Goal, No Goal) £ 2 (attribution: biological, psychological) £ 5 (age: 4year-olds, 5-year-olds, 7-year-olds, 10-year-olds, and adults) £ 4 (item: Video A, B, C,
and D) repeated-measures ANOVA was performed on the mean scores (see Fig. 2). As
indicated in the table of effects (Table 1), goal-directed agents elicited significantly more
attributions of biological and psychological capacities than aimless autonomous agents.
The presence of the goal affected the attributions of 5-year-olds through adults, but had no
effect on 4-year-olds (P ¼ 0:56). Biological capacities were also attributed more often
than psychological ones, and the difference between attributions of biological and psychological capacities was progressively larger at older ages. Although the difference in these
attributions increased with age, even 4-year-olds were more likely to attribute biological
capacities than psychological ones. Lastly, the presence of a goal affected adults’ biological attributions significantly more than their psychological attributions, but this distinctively biological effect of goal-directedness was attenuated in younger age groups (adults,
F ¼ 9:01; 10-year-olds, F ¼ 1:30; 7-year-olds, F ¼ 0:13; 5-year-olds, F ¼ 0:00; 4-yearolds, F ¼ 0:20).
To indicate which biological and psychological capacities goal-directedness affected, a
J.E. Opfer / Cognition 86 (2002) 97–122
109
Table 1
Table of effects: condition, attribution type, and item
Effect
F test
Post-hoc
Condition
Attribution type
Item
Age £ condition £ attribution type
Four-year-olds:
Attribution type
Five-year-olds:
Condition
Attribution type
Seven-year-olds:
Condition
Attribution type
Ten-year-olds:
Condition
Attribution type
Item £ condition
Adults:
Condition
Attribution type
Fð1; 150Þ ¼ 28:57***
Fð1; 150Þ ¼ 210:95***
Fð3; 450Þ ¼ 3:61*
Fð4; 150Þ ¼ 3:71**
Goal (M ¼ 3:68) . No Goal (M ¼ 2:28)
Biological (M ¼ 3:64) . Psychological (M ¼ 2:31)
Item C (M ¼ 3:2) . Item B (M ¼ 2:7)
Fð1; 30Þ ¼ 4:71*
Biological (M ¼ 2:96) . Psychological (M ¼ 2:56)
Fð1; 30Þ ¼ 5:14*
Fð1; 30Þ ¼ 13:75**
Goal (M ¼ 3:61) . No Goal (M ¼ 2:48)
Biological (M ¼ 3:17) . Psychological (M ¼ 2:48)
Fð1; 30Þ ¼ 8:66**
Fð1; 30Þ ¼ 17:93***
Goal (M ¼ 3:74) . No Goal (M ¼ 2:17)
Biological (M ¼ 3:38) . Psychological (M ¼ 2:54)
Fð1; 30Þ ¼ 8:45**
Fð1; 30Þ ¼ 103:36***
Fð3; 60Þ ¼ 3:7*
Goal (M ¼ 4:13) . No Goal (M ¼ 2:79)
Biological (M ¼ 3:38) . Psychological (M ¼ 2:54)
Goal (A ¼ B ¼ C ¼ D), No Goal (C, D . A, B)
Fð1; 30Þ ¼ 38:62****
Fð1; 30Þ ¼ 121:50****
Goal (M ¼ 3:90) . No Goal (M ¼ 1:84)
Biological (M ¼ 4:45) . Psychological (M ¼ 1:29)
Biological (Goal, M ¼ 5:91; No Goal, M ¼ 2:98),
Psychological (Goal, M ¼ 1:89; No Goal, M ¼ 0:69)
Goal (A ¼ B ¼ C ¼ D), No Goal (A, C . B, D)
Condition £ attribution type
Item £ condition
Fð3; 60Þ ¼ 9:01*
Fð1; 30Þ ¼ 3:96*
* P-value ¼ .05, ** p-value ¼ .01 *** p-value ¼ .001 **** p-value ¼ .0001.
Fig. 2. Attributions of biological and psychological capacities by age and condition.
110
J.E. Opfer / Cognition 86 (2002) 97–122
2 (condition: Goal, No Goal) £ 14 (attribution: die, grow, reproduce, need, need food,
need water, alive, want, pain, think, see, be happy, choose, and plan) repeated-measures
ANOVA was conducted on the scores (0–4) for each capacity separately. These comparisons are illustrated in Figs. 3–7. As indicated, 4-year-olds failed to distinguish goaldirected from autonomous agents in all of their attributions. In contrast, condition and
attribution interacted for every other age group (Ps , 0:05). Post-hoc comparisons using
the pooled variance term indicated that the presence of the goal affected the following at
the 0.05 level: 5-year-olds’ attributions of life and being able to grow, to need food, to see,
and to choose; 7-year-olds’ attributions of life and being able to grow, to need food, to
want, to think, and to plan; 10-year-olds’ attributions of life and being able to die, to need
something, to need water, to want, to think, and to see; and adults’ attributions of life and
being able to die, to grow, to need something, to reproduce, to need food and to need water,
to want, to choose, and to see.
Although the presence of the goal affected the attributions of both biological and
psychological capacities, generally only biological capacities were attributed to the
goal-directed agents at above-chance levels (P , 0:05). As illustrated in Figs. 5–9, 5year-olds were likely to attribute life; to which 7-year-olds added growth; to which fifth
graders added dying, needing something, and reproducing; to which adults finally added
needing food and needing water. Fifth graders’ attributions represented a singular exception to this trend: they were likely to say that even the aimless blobs “could make more
Fig. 3. Four-year-olds’ attributions of individual biological and psychological capacities by condition. Asterisks
indicate attributions that differ from chance (P , 0:05).
J.E. Opfer / Cognition 86 (2002) 97–122
111
Fig. 4. Five-year-olds’ attributions of biological and psychological capacities by condition. Asterisks indicate
attributions that differ from chance (P , 0:05).
little ones just like it”, and they were alone in attributing wanting to the goal-directed
blobs.
3.3. Correlations between attributions of self-movement and biological and psychological
capacities
To test whether judgments and perceptions of autonomous movement might independently elicit biological and psychological attributions, each participant’s self-movement
attributions (0–4) were regressed against his or her attributions of biological (0–7) and
psychological (0–7) capacities. In the No Goal condition, participants’ judgments of
autonomous movement were unrelated to their attributions of biological (rð79Þ ¼ 0:18,
n.s.) and psychological capacities (rð79Þ ¼ 0:08, n.s.). That is, when agents moved by
themselves but toward no goal, participants who said that those agents did in fact move by
themselves were no more likely to attribute biological and psychological capacities to the
agents than participants who did not judge the agents to move by themselves. Indeed, for
every age group that saw aimless self-movements, judgments of self-movement were
unrelated to biological and psychological attributions (R2 s ¼ 0:004–0:20, n.s.). In the
Goal condition, however, participants’ judgments of autonomous movement were strongly
correlated with their attributions of biological capacities (rð79Þ ¼ 0:51, P , 0:001),
though not psychological capacities (rð79Þ ¼ 0:00, n.s.), suggesting that the capacity for
self-movement, while thought to be correlated with other biological capacities, is not
thought to be sufficient to imply such capacities. Indeed, if we restrict our analysis of
112
J.E. Opfer / Cognition 86 (2002) 97–122
Fig. 5. Seven-year-olds’ attributions of biological and psychological capacities by condition. Asterisks indicate
attributions that differ from chance (P , 0:05).
variance only to those participants who judged all of the agents to move by themselves, we
still find that goal-directed agents are attributed about twice as many biological capacities
(M ¼ 5:09) as are the non-goal-directed agents (M ¼ 2:28) (Fð1; 34Þ ¼ 18:28,
P , 0:001), and about twice as many psychological capacities (Goal, M ¼ 2:71; No
Goal, M ¼ 1:15) (Fð1; 34Þ ¼ 4:72, P , 0:05) as well.
3.4. Agent identifications
To examine what the agents were thought to be, participants’ identifications of the four
agents were coded as either organismal (1) or inanimate (0). Organismal identifications
named any life form, including animal labels (e.g. “an animal”, “a jellyfish”), fanciful
organisms (e.g. “space creature”, “fish monster”), organ labels (e.g. “a stomach”, “a
heart”), collections of living things (e.g. “ten men”, “herd of animals”), and descriptions
of living things (e.g. “something that lives and can die”, “something like a dog”). Inanimate identifications named any non-living kind or artifact, including celestial bodies (e.g.
“shooting star”, “rocksteroid”), terrestrial substances (e.g. “rock”, “volcano guts”),
formerly living things (e.g. “ghost”), and artifacts (e.g. “magnet”, “spaceship”). Responses
that were uncodable, such as a persistent “I don’t know”, were eliminated, and the ratio of
organism identifications to the total number of codable identifications was calculated.
Interrater reliability on a randomly selected subset of responses (128/640) was 97%.
Each participant received a weighted score between 0 and 1, indicating the proportion of
J.E. Opfer / Cognition 86 (2002) 97–122
113
Fig. 6. Ten-year-olds’ attributions of biological and psychological capacities by condition. Asterisks indicate
attributions that differ from chance (P , 0:05).
agents identified as a living kind. A 2 (condition: Goal, No Goal) £ 5 (age: 4-year-olds, 5year-olds, 7-year-olds, 10-year-olds, and adults) ANOVA was conducted on these scores.
There was a main effect of condition (Fð1; 149Þ ¼ 28:35, P , 0:0001), indicating that
participants were more likely to identify the goal-directed agent (M ¼ 0:52) than the
autonomous agent (M ¼ 0:28) as an organism. Further, there was a main effect of age
(Fð4; 149Þ ¼ 8:02, P , 0:0001), and post-hoc tests indicated that adults (M ¼ 0:63)
provided more organismal identifications than any other age group (4-year-olds,
M ¼ 0:28; 5-year-olds, M ¼ 0:31; 7-year-olds, M ¼ 0:31; 10-year-olds, M ¼ 0:45;
Ps , 0:05), which did not differ. Condition and age did not interact (P ¼ 0:36). See
Fig. 8 for results.
Although 5-year-olds through adults were more likely to say that the goal-directed blob
was a life form than to say the autonomous blob was, their pattern of attributions suggest
that they did not consider the goal-directed blob to be the same kind of organism. As we
have seen, goal-directedness affected the children’s biological and psychological attributions about equally, whereas it primarily affected adults’ biological attributions. Possibly
children thought the goal-directed agent to be an animal (e.g. a bug or jellyfish), whereas
adults thought the goal-directed agent to be another sort of (non-psychological) organism
(e.g. a plant or germ or fungus).
To explore this possibility, participants’ organismal identifications of goal-directed
blobs were re-coded to separate animal–organismal identifications and non-animal–organismal identifications. Thus, every participant received two scores, one indicating the
114
J.E. Opfer / Cognition 86 (2002) 97–122
Fig. 7. Adults’ attributions of biological and psychological capacities by condition. Asterisks indicate attributions
that differ from chance (P , 0:05).
Fig. 8. Life form identifications by condition and age.
J.E. Opfer / Cognition 86 (2002) 97–122
115
proportion of identifications that named an animal (0–1) and another indicating the proportion that named another sort of living thing (0–1). To allow for the possibility that participants (presumably children) wanted to express that the thing was a non-animal organism,
but simply lacked names for such entities, “animal” identifications were defined in the
most strict sense, including only those life forms that are scientifically recognized to
belong to the animal kingdom (e.g. “baby”, “frog”, “insect”), whereas “other life form”
identifications included any other life form, including descriptions (e.g. “growing moving
thing”, “something chasing food”), fanciful organisms (e.g. “jelly monster”, “aliens”,
“space creature”), organs (“heart”, “stomach”), and plants and unicellular bodies (e.g.
“flower”, “germ”, “blood cell”, “bacteria”). This coding scheme thus allows children,
who were hypothesized to identify the blobs as animals, the greatest possible chance to
behave otherwise. Interrater reliability on a randomly selected subset of responses (128/
640) was 93%.
A 2 (life form: animal, other) £ 5 (age: 4-year-olds, 5-year-olds, 7-year-olds, 10-yearolds, and adults) ANOVA was conducted on these scores. There was a main effect of life
form (Fð1; 74Þ ¼ 8:89, P , 0:01), indicating that the goal-directed agent was most often
identified as an animal (M ¼ 0:33; other, M ¼ 0:19). Further, there was a main effect of
age (Fð4; 74Þ ¼ 6:17, P , 0:001); post-hoc tests indicated that adults were more likely
than any other age group to identify the goal-directed agent as some life form, and these
other age groups did not differ. Lastly, age and life form interacted (Fð4; 74Þ ¼ 9:97,
P , 0:0001), indicating that age affected other life form identifications (P , 0:001) but
failed to affect animal identifications (P ¼ 0:23). See Fig. 9 for results.
Fig. 9. Living kind identifications of goal-directed agents.
116
J.E. Opfer / Cognition 86 (2002) 97–122
4. Discussion
Movement provides important information about the identity of familiar animate entities. Numerous studies have demonstrated that adults can identify people – including their
sex and emotion – and other animals just by seeing how they move their joints (Brownlow,
Dixon, Egbert, & Radcliffe, 1997; Dittrich, Troscianko, Lea, & Morgan, 1996; Johansson,
1973; Mather & Murdoch, 1994; Montepare & Zebrowitz, 1993). Indeed, switching these
jointed movements for movements that violate the structure of the human form causes
even infants (and cats and pigeons) to take notice (Bertenthal, Proffitt, Spetner, & Thomas,
1985; Blake, 1993; Dittrich, Lea, Barrett, & Gurr, 1998; Fox & McDaniel, 1982; Ruff,
1985). That movement can convey critical information about an organism’s identity
cannot be disputed. However, what can be disputed is whether these movements (or
other sorts) are sufficient to identify living organisms more generally. After all, nonjointed organisms – worms, barnacles, plants, bacteria, fungi – are known to be alive,
so people must have discovered this fact without relying on the so-called “biological
motion” that is actually unique to organisms with legs. Moreover, the discovery of what
is alive is not just an historical problem, but an ongoing problem that faces young children
and adults of every generation whenever they encounter an unfamiliar entity.
The present results demonstrate that unfamiliar entities – blobs otherwise guessed to be
clouds, islands, and comets – are identified as living organisms when they are shown to be
capable of goal-directed movement. The goal-directedness appeared to be the decisive
factor in these displays: identical blobs that moved identically but toward no goal failed to
convince children and adults that they were living things. This biological interpretation of
goal-directed movement was in place as early as 5 years of age, when goal-directed
movement was first associated with life. From the age of 5 years to adulthood, the initial
judgment of life was successively elaborated with progressively more biological judgments and progressively fewer psychological ones, suggesting that goal-directed movement is more closely connected to the biological than psychological domain.
Although aimless autonomous agents were not convincing as living things, it was
important that the goal-directed agents were judged to have moved autonomously. If
goal-directed agents were thought to have been moved by an outside force (like an object
pulled or pushed toward a goal), participants were less likely to attribute biological
characteristics to the agent. Interestingly, though the movement was in fact identical,
goal-directed agents were more often judged to have moved autonomously than were
agents that did not move toward goals. However, even when aimless agents were thought
to have moved autonomously, they were not thought to have biological or psychological
properties.
These results support and delimit several claims about the role of movement in the
identification of animates (e.g. Gelman et al., 1995; Richards & Siegler, 1986; Scholl &
Tremoulet, 2000; Stewart, 1982). First, the results support the claim that some nonNewtonian movements – specifically, aimless autonomous movements – do not elicit
animacy judgments, much as Gelman et al. found when attempting to test the proposal
by Stewart (1982) along the same lines. In the present study, too, aimless autonomous
movements were ambiguous as animacy cues, even when participants were certain that the
movements were autonomous. As Csibra, Gergely and others (Csibra et al., 1999) have
J.E. Opfer / Cognition 86 (2002) 97–122
117
argued, the sort of mandatory interpretation of self-propulsion envisioned by some investigators (e.g. Baron-Cohen, 1994; Premack, 1990) would not be “a truly functional evolutionary adaptation as it would flood our perceptual system with false positives” (p. 263).
General skepticism about the ability to distinguish animates from inanimates on the
basis of motion, however, does not appear warranted. Some kinds of movement – specifically, goal-directed movements – reliably elicited animacy judgments, particularly life
judgments. This result is consistent with the proposal that animacy is directly perceived
(Scholl & Tremoulet, 2000), and it suggests that the critical cause of that perception is
goal-directed movement. Possibly, a mandatory biological interpretation of goal-directed
movement is more adaptive than a mandatory interpretation of autonomous movement
more broadly because goal-directed movement is a more conservative diagnostic of living
kinds (Binswanger, 1992; Mayr, 1982). Additionally, goal-directed movement might be
critical for identifying predators, whose biological goals are especially important to
discern: if you find yourself the goal of an unfamiliar animal’s contingent movement, it
may be safer to assume that its life depends on eating you than to assume that it really
wants to play with you.
Prior causal knowledge, such as that envisioned in the teleological core of preschoolers’
theory of biology (Keil, 1992; Opfer & Gelman, 2001) or psychology (Carey, 1985, 2000),
may also moderate the impact of goal-directed movement on animacy judgments. Consistent with this view, goal-directed movement did not signal life or any other biological or
psychological capacity to 4-year-olds, who are known to possess less biological knowledge than 5-year-olds (Laurendeau & Pinard, 1962), who did attribute life to the goaldirected agents. Although 4-year-olds were as likely as any age group to say that the goaldirected agents “moved so that [they] could get closer to [the goal]”, 1 they failed to
interpret the movement as a signal that the agents possessed any of the characteristics
of living and sentient kinds that were tested. Some investigators have proposed that goaldirectedness is interpreted psychologically by the age of 4 (Baron-Cohen, 1994; Premack
& Premack, 1996) or that intentional behavior qua intentional is directly perceived
(Dittrich & Lea, 1994). The current results do not support these proposals. Rather, they
support proposals (e.g. Csibra et al., 1999; Gergely et al., 1995) that young children
initially interpret goal-directed movement as goal-directed and can do so without agreeing
on any of the psychological or biological properties that the agent might possess.
Insufficient causal knowledge may have impacted 4-year-olds’ animacy judgments in at
least three ways. First, 4-year-olds may represent the distinction between goal-directed
and aimless autonomous movement, yet the distinction may not yet play a role in their
biological reasoning until they first appreciate how life is caused by goal-directed actions.
Csibra and Gergely (1998) have articulated a related argument concerning the elaboration
of teleological interpretations into causal mentalistic ones. The current results are consis1
The fact that 4-year-olds (or any age group) affirmed the question, “Did this move so that it could get closer to
this thing?” may not definitively confirm the impression of goal-directedness. The phrasing of the question may
not denote goal-directedness to the participants, especially if they do not differentiate the purpose construction
“so that it could” from the purpose-neutral conjunction “and”. Future work may identify the language that best
denotes the impression of goal-directedness, as well as the development of this knowledge, but as it stands, I
cannot say that the subjects’ impression of goal-directedness moderated the effect of the goal, only that the
presence of the goal elicited biological attributions.
118
J.E. Opfer / Cognition 86 (2002) 97–122
tent with their conclusions and suggest that they may apply to the biological domain as
well. Second, goal-directed movement may play an important role in 4-year-olds’ reasoning about living and sentient kinds, but the 4-year-olds did not spontaneously attempt to
find reasons for the agents’ movement as often as did older children: although 4-year-olds
often understand the causes of goal-directed events, explanatory questions are sometimes
needed to elicit evidence of this knowledge (Trabasso, Stein, Rodkin, & Munger, 1992).
Lastly, 4-year-olds may not differentiate life from such non-biological concepts as real
(Carey, 1985), thought that both the goal-directed and autonomous agents were equally
real, and thus failed to distinguish them with respect to life status and causally derivative
properties (such as growth and sentience). Each of these three alternatives posits that
causal knowledge plays an important role in children’s interpretation of goal-directed
movement, yet none need assume that the interpretation is mandatory.
Prior knowledge also seems to affect the meaning that older children and adults take
from goal-directed movement. Five-year-olds through adults, for example, seem to share
the belief that teleological agents are living things. Five-year-olds, for example, predict
that only animals will act teleologically (Opfer & Gelman, 2001) and believe that only
animals are living things (Carey, 1985; Hatano et al., 1993; Richards & Siegler, 1984); 10year-olds and adults, in contrast, predict that both plants and animals will act teleologically
(Opfer & Gelman, 2001) and typically believe that both plants and animals are living
things (Carey, 1985; Hatano et al., 1993; Richards & Siegler, 1984). Recent work suggests
that teaching 5-year-olds that both plants and animals act teleologically also convinces
them that plants are living things, whereas teaching them that plants and animals share
other biological properties (e.g. growth) fails to convince them that plants are living things
(Opfer & Siegler, 2001). Children’s judgments that novel goal-directed agents are alive
are consistent with these findings, suggesting that 5-year-olds through adults use goaldirected movement as a common measure for deciding whether an entity is alive. These
results are not consistent with the claim by Carey (1985, 2000) that 5-year-olds’ and
adults’ living things concept is incommensurable (literally, “without a common
measure”).
Differences in prior knowledge – specifically, in the range of organisms known to be
living things – may have led to differences in the attribution of psychological properties.
Adults’ knowledge about such insentient living things as germs and plants (Richards &
Siegler, 1984; Solomon & Cassimatis, 1999), for example, should lead them to consider
the possibility that an unfamiliar life form has many biological but no psychological
properties. Consistent with this hypothesis, many adults identified the goal-directed
agent (but not aimless agent) to be a microorganism, and the presence of the goal affected
their biological attributions more than their psychological ones. In contrast, younger
children may not believe that totally insentient organisms can exist (as Carey, 1985,
implied), or at least they do not consider them as a first hypothesis. Either of these
possibilities could explain the present results too: when children said that a goal-directed
agent was alive, they generally identified it to be a (sentient) animal of some type, and the
presence of the goal affected their biological and psychological attributions equally, with
the difference between these attributions increasing over time (see Coley, 1995, for a
similar observation). Against the former possibility, however, many children do seem to
believe that insentient organisms exist. Of the 32 participants in each age group, 12 5-year-
J.E. Opfer / Cognition 86 (2002) 97–122
119
olds, 13 7-year-olds, 16 10-year-olds, and 18 adults at least once attributed biological but
no psychological properties to an agent. (No participant, however, attributed psychological but no biological properties to any agent.) These results extend previous findings that
young children believe that some animals (such as houseflies) have little to nothing in the
way of a mental life (Gutheil, Vera, & Keil, 1998).
Given the range of living, goal-directed agents known by 10-year-olds (e.g. plants;
Opfer & Gelman, 2001; Richards & Siegler, 1984), it is surprising that they continued
to take the presence of the goal as an equally strong biological and psychological cue. One
possibility is that 10-year-olds believe that only plants and animals are alive and simply do
not know that other life forms exist. Thus, when they saw a motion that indicated life to
them, their task was to decide whether the unfamiliar life form was a plant or an animal.
Given that the blob was neither rooted nor leafy nor green, they could have deduced that
all the life form alternatives had been exhausted, and thus it was likely that the life form
was an animal (i.e. a psychological organism). At the very least, the present data suggest
that this is not the whole story. First, nine of 32 10-year-olds identified the agent as a nonplant, non-animal, microscopic living thing (e.g. “blood cell”, “bacteria”, “germ”). Thus,
at least some 10-year-olds know that plants and animals are not the only life forms. If we
additionally count the number of 10-year-olds who guessed the agent to be part of an
organism (e.g. “brain”, “stomach”, “roots”), half named non-animal, non-plant life forms.
Clearly, for many 10-year-olds, living things is broader than plant or animal. Therefore,
the fact that the blob did not look like a plant cannot wholly account for their performance.
Another possibility is that goal-directed action affects biological and psychological
inferences equally for all ages unless overridden by extra knowledge about the agents
themselves (e.g. that plants and microorganisms are insentient). This possibility maintains
a prerequisite connection between biology and goal-directedness, credits children with
knowing that insentient life forms can exist, and can account for 10-year-olds’ performance. Additionally, it yields two specific predictions. First, 5- through 10-year-olds
could behave as adults did just in case they learn that microscopic life forms and plants
are insentient. Second, if adults cannot positively identify a goal-directed agent as a plant
or microscopic life form (e.g. triangles in Heider & Simmel, 1944), even their attributions
may resemble those of 5- through 10-year-olds: that is, blocking all information about the
agent itself, goal-directed action may serve as an equally strong biological and psychological cue.
In sum, the results support the proposal that 5-year-olds through 10-year-olds and adults
share at least one standard in deciding what is “alive”, that is, to be “alive” is to be an agent
capable of goal-directed action (animals for 5-year-olds, animals and other organisms for
10-year-olds and adults). Accordingly, children and adults can identify living things from
goal-directed motion alone. However, children and adults may differ in the sorts of
psychological capacities they think to correlate with moving toward goals: for children,
goal-directedness affected attributions of biological and psychological capacities fairly
similarly, whereas for adults, it affected attributions of biological capacities much more
than psychological ones. These age differences in biology and psychology judgments
suggest that goal-directed action could be an early standard for identifying living things,
but that other information may play a role in psychology judgments.
120
J.E. Opfer / Cognition 86 (2002) 97–122
Acknowledgements
This study was funded in part by NIMH Grant No. T32MH19102. Portions of this study
were presented at the biennial meeting of the Society for Research in Child Development,
April 2001, Minneapolis, MN, and the biennial meeting of the Cognitive Development
Society, October 1999, Chapel Hill, NC. The author would like to thank Susan Gelman,
Marilyn Shatz, Henry Wellman, Paul Pintrich, Robert Siegler, David Rakison, Brian
Scholl, and two anonymous reviewers for their valuable comments on earlier drafts,
Linda Opfer at Forest Road Elementary School, and the faculty and staff at the University
of Michigan Children’s Center and the Children’s Center for Working Families.
References
Bassili, J. N. (1976). Temporal and spatial contingencies in the perception of social events. Journal of Personality
and Social Psychology, 33, 680–685.
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind. Cambridge, MA: MIT Press.
Bertenthal, B. I., Proffitt, D. R., Spetner, N. B., & Thomas, M. A. (1985). The development of infant sensitivity to
biomechanical motions. Child Development, 56, 531–543.
Binswanger, H. (1992). Life-based teleology and the foundation of ethics. The Monist, 75, 84–103.
Blake, R. (1993). Cats perceive biological motion. Psychological Science, 4, 54–57.
Brownlow, S., Dixon, A. R., Egbert, C. A., & Radcliffe, R. D. (1997). Perception of movement and dancer
characteristics from point-light displays of dance. Psychological Record, 47, 411–421.
Bullock, M. (1985). Animism in childhood thinking: a new look at an old question. Developmental Psychology,
21, 217–225.
Carey, S. (1985). Conceptual change in childhood. Cambridge, MA: MIT Press.
Carey, S. (1995). On the origins of causal understanding. In S. Sperber, D. Premack & A. Premack (Eds.), Causal
cognition: a multi-disciplinary debate (pp. 268–308). Cambridge: Oxford University Press.
Carey, S. (2000). Sources of conceptual change. In E. K. Scholnick, K. Nelson, S. A. Gelman & P. H. Miller
(Eds.), Conceptual development: Piaget’s legacy (pp. 293–326). Mahwah, NJ: Erlbaum.
Coley, J. (1995). Emerging differentiation of folkbiology and folkpsychology: attributions of biological and
psychological properties to living things. Child Development, 66, 1856–1874.
Csibra, G., & Gergely, G. (1998). The teleological origins of mentalistic action explanations: a developmental
hypothesis. Developmental Science, 1, 255–259.
Csibra, G., Gergely, G., Biro, S., Koos, O., & Brockbank, M. (1999). Goal attribution without agency cues: the
perception of “pure reason” in infancy. Cognition, 72, 237–267.
Dittrich, W. H., & Lea, S. E. G. (1994). Visual perception of intentional motion. Perception, 23, 253–268.
Dittrich, W. H., Lea, S. E. G., Barrett, J., & Gurr, P. R. (1998). Categorization of natural movements by pigeons:
visual concept discrimination and biological motion. Journal of Experimental Analysis of Behavior, 70, 281–
299.
Dittrich, W. H., Troscianko, T., Lea, S. E. G., & Morgan, D. (1996). Perception of emotion from dynamic pointlight displays represented in dance. Perception, 25, 727–738.
Dolgin, K., & Behrend, D. A. (1984). Children’s knowledge about animates and inanimates. Child Development,
55, 1546–1650.
Fox, R., & McDaniel, C. (1982). The perception of biological motion by human infants. Science, 218, 486–487.
Gelman, R., Durgin, F., & Kaufman, L. (1995). Distinguishing between animates and inanimates: not by motion
alone. In S. Sperber, D. Premack & A. Premack, Causal cognition: a multi-disciplinary debate. Cambridge:
Oxford University Press.
Gelman, S. A., & Opfer, J. E. (in press). Development of the animate–inanimate distinction. In U. Goswami (Ed.),
Handbook of childhood cognitive development. Oxford: Blackwell.
J.E. Opfer / Cognition 86 (2002) 97–122
121
Gergely, G., & Csibra, G. (1997). Teleological reasoning in infancy: the infant’s naive theory of rational action. A
reply to Premack and Premack. Cognition, 63, 227–233.
Gergely, G., Nadasdy, Z., Csibra, G., & Biro, S. (1995). Taking the intentional stance at 12 months of age.
Cognition, 56, 165–193.
Gutheil, G., Vera, A., & Keil, F. C. (1998). Do houseflies think? Patterns of induction and biological beliefs in
development. Cognition, 66, 33–49.
Hatano, G., Siegler, R. S., Richards, D. D., Inagaki, K., Stavy, R., & Wax, N. (1993). The development of
biological knowledge: a multi-national study. Cognitive Development, 8, 47–62.
Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. American Journal of Psychology,
57, 243–249.
Johansson, G. (1973). Visual perception of biological motion and a model for its analysis. Perception and
Psychophysics, 14, 201–211.
Keil, F. C. (1992). The origins of an autonomous biology. In M. A. Gunnar & M. Maratsos (Eds.), Minnesota
Symposium on Child Psychology, Vol. 25. Hillsdale, NJ: Erlbaum.
Laurendeau, M., & Pinard, A. (1962). Causal thinking in the child: a genetic and experimental approach. New
York: International Universities Press.
Looft, W. R., & Bartz, W. H. (1969). Animism revisited. Psychological Bulletin, 71, 1–19.
Mandler, J. M. (1992). How to build a baby II. Conceptual primitives. Psychological Review, 99, 587–604.
Mather, G., & Murdoch, L. (1994). Gender discrimination in biological motion displays based on dynamic cues.
Proceedings of the Royal Society of London, 258 (1353), 273–279.
Mayr, E. (1982). The growth of biological thought. Cambridge, MA: Harvard University Press.
Michotte, A. (1963). The perception of causality. New York: Basic Books.
Montepare, J. M., & Zebrowitz, L. A. (1993). A cross-cultural comparison of impressions created by age-related
variations in gait. Journal of Nonverbal Behavior, 17, 55–68.
Montgomery, D. E. (1996). The role of action-initiation in young children’s causal explanations of action.
Cognitive Development, 11, 467–489.
Opfer, J. E., & Gelman, S. A. (2001). Children’s and adults’ models for predicting teleological action: the
development of a biology-based model. Child Development, 72, 1367–1381.
Opfer, J. E., & Siegler, R. S. (2001, October). Teleological action speaks louder than words: a microgenetic
analysis of conceptual change in naive biology. Poster presented at the biennial meeting of the Cognitive
Development Society, Virginia Beach, VA.
Piaget, J. (1929). The child’s conception of the world, London: Routledge and Kegan Paul.
Poulin-Dubois, D., & Héroux, G. (1994). Movement and children’s attributions of life properties. International
Journal of Behavioral Development, 17, 329–347.
Poulin-Dubois, D., Lepage, A., & Ferland, D. (1996). Infants’ concept of animacy. Cognitive Development, 11,
19–36.
Premack, D. (1990). The infant’s theory of self-propelled objects. Cognition, 36, 1–16.
Premack, D., & Premack, A. J. (1996). Intention as psychological cause. In S. Sperber, D. Premack & A. Premack
(Eds.), Causal cognition: a multi-disciplinary debate. Cambridge: Oxford University Press.
Premack, D., & Premack, A. J. (1997). Infants attribute value ^ to the goal-directed actions of self-propelled
objects. Journal of Cognitive Neuroscience, 9, 848–856.
Reed, E. S. (1997). Why perceiving produces ontological knowledge. Human Development, 40, 245–253.
Richards, D., & Siegler, R. S. (1984). The effects of task requirements on children’s life judgments. Child
Development, 55, 1687–1696.
Richards, D., & Siegler, R. S. (1986). Children’s understandings of the attributes of life. Journal of Experimental
Child Psychology, 42, 1–22.
Ruff, H. A. (1985). Detection of information specifying the motion of objects by 3- and 5-month-old infants.
Developmental Psychology, 21, 295–305.
Scholl, B. J., & Tremoulet, P. D. (2000). Perceptual causality and animacy. Trends in Cognitive Sciences, 4, 299–
309.
Solomon, G. E. A., & Cassimatis, N. L. (1999). On facts and conceptual systems: young children’s integration of
their understanding of germs and contagion. Developmental Psychology, 35, 113–126.
Stewart, J. A. (1982). Perception of animacy. Unpublished dissertation, University of Pennsylvania, Philadelphia.
122
J.E. Opfer / Cognition 86 (2002) 97–122
Stewart, J. A. (1984). Object motion and the perception of animacy. Paper presented at the meeting of the
Psychonomic Society, San Antonio, TX.
Trabasso, T., Stein, N. L., Rodkin, P. C., & Munger, M. P. (1992). Knowledge of goals and plans in the on-line
narration of events. Cognitive Development, 7, 133–170.
White, P. A., & Milne, A. (1997). Phenomenal causality: impressions of pulling in the visual perception of objects
in motion. American Journal of Psychology, 110, 573–602.